Synthetic health data in Canada: A scoping review of methods, applications, and data sources

Comprehensive Summary

This research presented by Golandouz et al. performed a scoping review studying trends of health across Canada through synthetic data as a way to preserve patient confidentiality. The scoping review involved analyzing all available research studies up until September 2024. The studies were reviewed base on the type of health data, the source, methods, and evaluations of quality. Results showed that only 12 articles (out of 232) adequately fit the criteria being peer-reviewed and grey literature. Machine learning predictive and generative models as well as sampling approach were also used to develop synthetic data. Golandouz et al. acknowledged the need for higher consistent quality synthetic data and that available studies were largely survey based.

Outcomes and Implications

Synthetic data covering health databases could be a revolutionary tool expanding the reach of Canadian research across provinces and in cross-jurisdictional studies.​​ The use of synthetic data as an alternative protecting patient information for collaboration across provinces in Canada could improve potentially research speed and equity. Not only that, but synthetic data allows for artificial intelligence to be trained without needing direct patient exposure.